摘要 |
A spoken word (a) comprises several voiced and silent periods (b), whilst a recognition process (c) comprises the data acquisition of the head of a word (A), pre-selection (B), main selection (C) and the output of the result (D). When an unknown voice turns into a voiceless condition (b-a), the pre-selection process (B) is interrupted and a main selection process (C) takes place. However, if another voiced condition is established within the word, (b-b) the current main selection process (C) is aborted and the pre-selection process (B) is presumed. At the end of the word, the result (D) of the main selection is outputted. |